Leonard Teo: Tell us about the production process and how Eternal Gaze took shape from concept to finished render/composite.
Sam Chen: Having learned from past personal mistakes, I learned not to rush into production too hastily and to delay touching the computer for as long as possible until the script and the hand-drawn animatic were pretty solid. I tried to spend as much effort on the screenwriting phase, or more accurately, the storyboarding and re-storyboarding phase for as long as my patience could tolerate. Only then would jumping on the computer make any sense. Like many animation projects at this point, it's just a matter of refining the shots taking each from pencil sketches, to rough layout, blocking animation, final animation, and eventually rendering and compositing.
So with 220 shots slated over a running time of 16 minutes, on any given day, I could be working on 5 shots at a time with each in various phases of the pipeline. As each shot completed a phase, it was dropped into the non-linear editor timeline to check for flow, continuity, and emotional impact. So it's a constant in-out, microscopic-macroscopic type of workflow with the occasional popping up to the highest level to make sure the "Big Picture" was still resonating ultimately. It certainly helped to be schizophrenic too.
Leonard Teo: As the film is B+W, did that ease things up on the texturing/lighting/ composition?
Sam Chen: Working in black & white actually simplified a few things while introducing new issues at the same time. For example, the human flesh tone is one of the most challenging things to reproduce correctly in CG. But with b&w, it's less of an issue. This freed me up to focus more on composition and form without being distracted by colors. The result is hopefully a visually stronger film.
In achieving the 3 very different lighting schemes required, this is where my practical experience in traditional b&w photography came in handy. Act 1 has a neutral look while the lighting quickly turned dark and menacing in Act 2. In Act 3, it's ethereal and heavenly. I was pleasantly surprised at how well CG was able to mimick the b&w filmic look that I was after. It is my hope that others will push CGI in the direction of b&w even more and come up with some amazingly looking films.
Leonard Teo: Tell us about modelling, animating and rendering Alberto. From his basic model, to rigging to his hair.
Sam Chen: The modeling of the Giacometti character really started from his head. Believe it or not, it's basically nothing but a NURBS sphere with lots of CV pushing and pulling, along with the strategic insertion of isoparms here and there. I chose to do most things in NURBS because at the time I embarked on modeling the main character, subdivision surfaces were still not reliable and efficient enough. Ultimately, Giacometti was modeled with a combination of NURBS, Polygons, and SubD Surfaces depending on the strengths and weaknesses of each paradigm.
In rigging Giacometti, only his legs had IK while the arms used only FK. There was no cloth simulation involved so any cloth-like flapping were animated by hand using bones that were rigged within for instance, his jacket and pants. As for the face, instead of using Morph Targets, a bone-based system was used to simulate the contraction and expansion of facial muscles.
So depending on what expressions or emotions were required, I spent much effort mastering and understanding the interesting relationship between human emotions and which facial muscles were responsible for them.
For animation, a hybrid pose-to-pose method was used where after the key poses were set, there was still enough room for straight ahead spontaneity. Some of the best animation moments involved "happy accidents." I tried hard to recognize these serendipitous moments and let them happen organically. That's when animation becomes adventurous and the most fun. By the way, 100-percent of the animation in "Eternal Gaze" was hand-keyframed. No mocap and simulaton were ever involved.
For rendering, scenes were often separated and rendered into multiple layers for better control of lighting, and depth-of-field. Depending on the scene, there could be a separate layer for background, character, hair, sculpture, shadows, smoke, rain, and extreme foreground elements. Area Lights were used as much as possible as they tend to look more natural with softer highlights and controllable dimensions. Hair was always rendered on separate passes and composited onto the character layer in AfterEffects. The final step usually involved precise control of depth-of-field, selective defocusing, layered glows, and final value adjustments. [CGN]
Timeframe: 2.5 years of production, 0.5 years of
research and story development.
Software: Photoshop for textures; Maya for 3D; AfterEffects for post; and Premiere for editing.
Hardware: Various Windows PC's including a Sony VAIO
laptop. Canopus DVStorm for video output.
Credited artists: Sam Chen (screenwriter, director,
producer, animator, modeler, editor, technical
director, designer); Jamey Scott (music composer,
sound designer, co-producer)
Coffee consumed: 500 gallons and counting...!
Leonard Teo is the Editor of CGNetworks, email him here. Eternal Gaze will be showing at the SIGGRAPH 2003 Electronic Theater, 26-31 July 2003, San Diego, California. For more information about Eternal Gaze, visit the official website at www.EternalGaze.com
Discuss this article on CGTalk.com